Study on Improved ODDDP Algorithm for Cascade Reservoirs Optimal Operation
The problems of curse of dimensionality and easily falling into local optimum are the hot and difficult points in solving the model of cascade reservoirs optimal operation.The problem of curse of dimensionality for Discrete Differen-tial Dynamic Programming algorithm(DDDP)in solving the problem of cascade reservoirs optimal operation was solved by Orthogonal Discrete Differential Dynamic Programming algorithm(ODDDP).However,the ODDDP does not over-come the defects of relying on the initial solution quality and easily falling into local optimum.By introducing Gaussian random variables,the M-IWO-ODDDP was proposed by referencing the spatial diffusion mechanism of the invasive weed algorithm and the periodicity of the trigonometric cosine function.The global convergence of the algorithm can be im-proved by improving the global search ability and search depth of the ODDDP.The convergence and robustness of the M-IWO-ODDDP were tested by using Schaffer and Shubert functions.The 11 cascade reservoirs in Wujiang River Basin was used as a case study.The results show that the optimization effect of M-IWO-ODDDP is significant and it not only has no obvious dependence on the quality of the initial solution,but also has good robustness.
万方数据
维普
